CN115931783A - GEE-based large-scale region lake and reservoir chlorophyll a mapping method - Google Patents
GEE-based large-scale region lake and reservoir chlorophyll a mapping method Download PDFInfo
- Publication number
- CN115931783A CN115931783A CN202210710949.5A CN202210710949A CN115931783A CN 115931783 A CN115931783 A CN 115931783A CN 202210710949 A CN202210710949 A CN 202210710949A CN 115931783 A CN115931783 A CN 115931783A
- Authority
- CN
- China
- Prior art keywords
- chlorophyll
- lake
- reservoir
- data
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The invention discloses a GEE-based large-scale regional lake and reservoir chlorophyll a mapping method, and relates to a regional lake and reservoir chlorophyll a mapping method. The invention provides a GEE-based large-scale regional lake reservoir chlorophyll a mapping method, which can accurately estimate multi-space-time scale chlorophyll a concentration and space change condition monitoring thereof. The specific method comprises the following steps: 1. acquiring actual measurement data of chlorophyll a; 2. acquiring remote sensing image data; 3. screening out effective matching data; 4. carrying out random distribution on the effective matching data; 5. constructing a lake and reservoir chlorophyll a random forest inversion model; 6. and applying the chlorophyll a inversion model obtained in the fifth step to the zenith angle reflectivity stored by GEE to realize rapid mapping of chlorophyll a in the lake reservoir in large-scale areas and multiple space-time scales. The method constructs the large-scale regional lake and reservoir chlorophyll a inversion model suitable for zenith angle reflectivity, and can realize multi-scale time sequence and space water quality and eutrophication degree monitoring and rapid mapping.
Description
Technical Field
The invention relates to a regional lake reservoir chlorophyll a mapping method.
Background
Chlorophyll a (Chlorophyl-a, chl-a) is one of important indexes for monitoring lake and reservoir water environments, and the concentration change of the Chlorophyll a can comprehensively reflect the water quality condition. The water quality deterioration, dissolved oxygen reduction, algal bloom outbreak and algal toxin release are easily caused by the overhigh content of the chlorophyll a in the water body, the living environment of aquatic organisms is seriously threatened, and the safety of water for production and living of human beings is seriously threatened. Under the conditions of severe human activity influence and global warming, the situation of lake and reservoir eutrophication becomes increasingly serious. According to statistics, the global area is more than 25km 2 Of the 2058 lakes, 63.1% were in an eutrophic state, with asia beginning with 54% eutrophic lake sites. Thus, lake eutrophication has become a global water quality problem affecting most water ecosystems.
The traditional water quality monitoring is a field sampling-indoor chemical analysis method, and is also a main method adopted by the environmental and ecological environment monitoring department in China at present, although the method has high precision, the method has more defects: firstly, a single sampling point cannot replace the water quality condition of the whole lake and reservoir and cannot comprehensively reflect the scale pollution condition of a drainage basin; secondly, a large amount of manpower, material resources and financial resources are consumed in the sampling and experiment process; furthermore, visual sceneries and continuous spatio-temporal sequence expression cannot be realized. Particularly, the study on the time-space change of Chl-a in a single lake reservoir is limited in the aspect of monitoring study in a large scale range, and the requirement of the current society on water quality monitoring is difficult to meet.
The aim of monitoring and researching the Chl-a concentration of large-scale lakes and reservoirs is developed, firstly, the nutrition conditions of local-scale (single lake reservoir) or multi-scale (regional lake and reservoir groups) lakes and reservoirs are controlled for different environmental factors, the eutrophication state of individual lakes and reservoirs is mainly controlled by the biological factors in the lakes and reservoirs, and the eutrophication mechanism process of lakes and reservoirs can be known through fixed-point in-situ sampling of the lakes and reservoirs. In contrast, abiotic factors (i.e., climatic and hydrological factors) and their associations are key factors in determining the biogeochemistry of lakes on multiple scales, which often require monitoring tens or hundreds of lake reservoirs to establish a spatio-temporal pattern of nutritional status. Therefore, in order to understand the response mechanism of the lakes and reservoirs to global climate and regional environmental changes, large-scale research on the nutritional status of the lakes and reservoirs is required.
The remote sensing technology has the advantages of real-time rapidness, accuracy and low cost, can acquire satellite archived data to realize continuous time-space change analysis, provides a trigger for national, intercontinental or global large-scale water quality monitoring and evaluation, can realize continuous and rapid monitoring and evaluation on lakes and reservoirs by adopting an easy analysis index and mode, and better provides a reference basis for implementation of water resource management and pollution control policies. But the method is limited by large regional space difference and computer processing capability, and a national, intercontinental or global lake reservoir chlorophyll a concentration synchronous satellite remote sensing monitoring algorithm is not reported temporarily, and a large-scale regional chlorophyll a rapid mapping method is not applied.
Disclosure of Invention
The invention provides a GEE-based large-scale regional lake reservoir chlorophyll a mapping method, which can accurately estimate multi-space-time scale chlorophyll a concentration and space change condition monitoring thereof.
The large-scale region lake and reservoir chlorophyll a mapping method based on GEE is carried out according to the following steps: 1. collecting lake and reservoir water body samples with different climatic and geographical environment backgrounds to obtain actual measurement data of chlorophyll a;
2. acquiring remote sensing image data corresponding to a water sample collected on site based on a GEE computing cloud platform, and extracting zenith reflectivity values and remark information of each wave band;
3. removing remote sensing image data of clouds, mountain shadows and atmospheric aerosol according to remark information, and screening effective matching data;
4. randomly distributing the effective matching data obtained in the third step, wherein 70% of the effective matching data are chlorophyll a inversion model correction data sets, and 30% of the effective matching data are model verification data sets;
5. constructing a lake-reservoir chlorophyll a random forest inversion model based on chlorophyll a measured data, zenith reflectance values, a chlorophyll a inversion model correction data set and a chlorophyll a inversion model verification data set, and specifically comprising the following steps:
a) Analyzing the correlation between the zenith reflectance value of each waveband and actually-measured chlorophyll a data aiming at the model correction data set, and selecting a waveband with high correlation;
b) Aiming at the model correction data set, training different combination forms of wave bands with good correlation with chlorophyll a concentration, and taking the different combination forms and the chlorophyll a concentration value as model input variables;
c) Parameterizing the model correction data set to construct a chlorophyll a remote sensing inversion model;
d) Verifying a chlorophyll a remote sensing inversion model aiming at the model verification data set;
6. and applying the chlorophyll a inversion model in the fifth step to the zenith reflectivity of the remote sensing image on the GEE platform to realize rapid drawing of chlorophyll a in the lake reservoir in a large-scale area and multiple space-time scales.
In the first step, an acetone extraction method is adopted in the method for measuring the concentration of the chlorophyll a in the water body, and the concentration value (unit, mu g/L) of Chl-a is calculated according to a Jeffrey-Humphrey equation and the volume of a filtered water sample.
In the second step, the remote sensing data directly selects and uses the zenith reflectivity (TOA) data of Landsat series satellites (TM/ETM +/OLI) stored by GEE.
Further, in the second step, the remote sensing data acquisition time window is +/-7 days, the space window is 3 x 3 pixels, and the process can be realized through the programming service of the GEE cloud platform.
In the third step, in the process of acquiring the effective matching data, according to remark information, removing remote sensing image data which is greatly influenced by cloud, mountain shadow, atmospheric aerosol and the like, and screening out the effective matching data;
in the fourth step, the effective matching data is randomly divided into chlorophyll by using a random function randa inverse model correction dataset (N) Correction of ＝2/3×N General assembly 70%) and model validation data set (N) Checking the right ＝1/3×N General (1) ，30％)；N General assembly Representing valid match data, N Correction of Data set of correction data representing chlorophyll a inverse model, N Correction by testing Representing a chlorophyll a inverse model validation dataset.
And in the fifth step, performing high correlation test of zenith reflectance values of all wave bands and actually-measured chlorophyll a by adopting correlation analysis, and preferably selecting the wave band with better correlation with the chlorophyll a as an input variable of a random forest algorithm to construct an inversion model. The sensor band corresponding to the zenith reflectivity (TOA) of Landsat satellites (TM/ETM +/OLI) is shown in Table 1, and the test of the high correlation between the zenith reflectivity and the actual chlorophyll a in each band is shown in Table 2.
TABLE 1 response sensor band
| Landsat series sensor | Red band (wavelength) | Blue wave band (wavelength) |
| TM | B3(0.63-0.69μm) | Band 1(0.45-0.52μm) |
| ETM+ | B3(0.63-0.69μm) | Band 1(0.45-0.52μm) |
| OLI | B4(0.64-0.67μm) | Band3(0.45-0.51μm) |
TABLE 2 correlation test of band combinations
B represents a wavelength band, G represents a red wavelength band, R represents a red wavelength band, and NIR represents a red wavelength band
And in the fifth step, verifying the concentration of the chlorophyll a of the lake and reservoir pixels constructed in the fifth step through a model verification data set.
As a further improvement of the invention, the scheme comprises the step of applying the constructed chlorophyll a inversion model to the zenith remote sensing reflectivity of Chinese Landsat series satellite remote sensing data stored in GEE to realize the lake (area) in the North China in 1985-2020>1km 2 ) The chlorophyll a concentration change is rapidly mapped.
Aiming at the concentration of lake chlorophyll a, the lake chlorophyll a content is phytoplankton cytochrome chlorophyll a, an inversion model of large-scale area lake and reservoir chlorophyll a suitable for zenith reflectivity is constructed by combining continuous ten-year synchronous chlorophyll a actual measurement data and zenith reflectivity on the basis of remote sensing big data and a GEE computer cloud platform, and a machine learning random forest algorithm is adopted. The inversion model can realize multi-scale time sequence and space water quality and eutrophication degree monitoring and rapid mapping, has good model effect of estimating the chlorophyll a of the lake reservoir, is feasible through the verification of an actual measurement sample point model, and provides data support for the evaluation and comprehensive comparison of the large-scale regional lake reservoir water quality.
Drawings
FIG. 1 is a schematic diagram showing a distribution diagram of the investigation of lakes and reservoirs in FIG. 1;
FIG. 2 is a diagram showing the correlation coefficient of the zenith refractive index and the actually measured chlorophyll-a concentration of the Landsat series of the satellite in example 1;
FIG. 3 is an inversion and actual measurement fitting graph of the random forest verification model for chlorophyll a in the water body in the lake and reservoir in example 1;
FIG. 4 is an inversion and actual measurement fitting graph of the random forest verification model for chlorophyll a in the water body in the lake and reservoir in example 1;
FIG. 5 shows the lake repository in northern China in 1985 in example 1 (>1km 2 ) Chlorophyll a concentration profile;
FIG. 6 shows the lake repository in northern China in 1990 of example 1 (>1km 2 ) Chlorophyll a concentration profile;
FIG. 7 shows northern lake deposits (in 1995) in northern China in example 1>1km 2 ) Chlorophyll a concentration profile;
FIG. 8 shows the lake in northern China in 2000 of example 1>1km 2 ) Chlorophyll a concentration profile;
FIG. 9 shows the 2005 northern lake repository in China (example 1)>1km 2 ) Chlorophyll a concentration profile;
FIG. 10 shows the 2010 northern lake repository in China in example 1>1km 2 ) Chlorophyll a concentration profile;
FIG. 11 shows the lake in northern China lake (2015) in example 1>1km 2 ) Chlorophyll a concentration profile;
FIG. 12 shows the northern lake repository in China in 2020 of example 1>1km 2 ) Chlorophyll a concentration profile;
Detailed Description
The first specific implementation way is as follows: the large-scale region lake and reservoir chlorophyll a mapping method based on GEE is carried out according to the following steps:
1. collecting lake and reservoir water body samples with different climatic and geographical environment backgrounds to obtain actual measurement data of chlorophyll a;
2. remote sensing image data corresponding to a field-collected water sample are obtained based on a GEE computing cloud platform, and zenith reflectivity values and remark information of all wave bands are extracted;
3. removing remote sensing image data of clouds, mountain shadows and atmospheric aerosol according to remark information, and screening effective matching data;
4. randomly distributing the effective matching data obtained in the third step, wherein 70% of the effective matching data are chlorophyll a inversion model correction data sets, and 30% of the effective matching data are model verification data sets;
5. constructing a lake-reservoir chlorophyll a random forest inversion model based on chlorophyll a measured data, zenith reflectance values, a chlorophyll a inversion model correction data set and a chlorophyll a inversion model verification data set, and specifically comprising the following steps:
a) Aiming at the model correction data set, analyzing the correlation between the zenith reflectance value of each wave band and actually measured chlorophyll a data, and selecting a wave band with high correlation;
b) Aiming at the model correction data set, training different combination forms of wave bands with good correlation with chlorophyll a concentration, and using the different combination forms and the chlorophyll a concentration value as model input variables;
c) Parameterizing the model correction data set to construct a chlorophyll a remote sensing inversion model;
d) Verifying a chlorophyll a remote sensing inversion model aiming at the model verification data set;
6. and applying the chlorophyll a inversion model in the fifth step to the zenith reflectivity of the remote sensing image on the GEE platform to realize rapid drawing of chlorophyll a in the lake reservoir in a large-scale area and multiple space-time scales.
In the second step of the embodiment, the remote sensing data directly selects and uses the zenith reflectivity (TOA) data of Landsat series satellites (TM/ETM +/OLI) stored by GEE; further, in the second step, the time window of remote sensing data acquisition is +/-7 days, the space window is 3 x 3 pixels, and the process can be realized through GEE cloud platform programming business.
In the third step of the embodiment, in the process of acquiring the effective matching data, remote sensing image data greatly influenced by cloud, mountain shadow, atmospheric aerosol and the like is removed according to remark information, and the effective matching data is screened out.
In step five of the implementation method, correlation analysis is adopted to carry out high correlation test between zenith reflectance values of all wave bands and actually-measured chlorophyll a, and the wave band with better correlation with the chlorophyll a is preferably used as an input variable of a random forest algorithm to construct an inversion model.
In the fifth step of the embodiment, the concentration of the chlorophyll a in the lake and reservoir pixels constructed in the fifth step is verified through the model verification data set.
As a further improvement of this embodiment, the methodThe method comprises the step of applying the constructed chlorophyll a inversion model to the zenith remote sensing reflectivity of the remote sensing data of the Chinese Landsat series satellite stored in GEE to realize the lake (area) in the North China in 1985-2020>1km 2 ) The chlorophyll a concentration change is rapidly mapped.
The second embodiment is as follows: the first difference between the present embodiment and the specific embodiment is: the method for actually measuring the data of the chlorophyll a in the first step adopts an acetone extraction method, namely, the concentration value of the chlorophyll a is calculated according to a Jeffrey-Humphrey equation and the volume of a filtered water sample, and the unit is mu g/L. Other steps and parameters are the same as those in the first embodiment.
The third concrete implementation mode: the first difference between the present embodiment and the specific embodiment is: in the third step, the random function rand in Matlab software is used for randomly dividing the effective matching data into 70 percent chlorophyll a inversion model correction data sets, N Correction of ＝2/3×N General assembly (ii) a 30% of the model validation data set, N Checking the right ＝1/3×N General assembly (ii) a Wherein N is General assembly Representing valid match data, N Correction of Representing chlorophyll a inverse model correction data set, N Checking the right Representing a chlorophyll a inverse model validation dataset. Other steps and parameters are the same as those in the first embodiment.
Example 1 in this example 1, the area of the north area (north of Qinling mountain and Huaihe river) in China is more than 1km 2 The lake is taken as an object, and the drawing method of chlorophyll a in large-scale regional lakes and reservoirs based on GEE is adopted to carry out the drawing of chlorophyll a in regional lakes and reservoirs. The method comprises the following specific steps:
1. collecting a lake and reservoir water body sample, and obtaining actual measurement data of chlorophyll a. The lake reservoirs in China are numerous, and the climate partitions, geographic landscape units and drainage basins of different lake reservoirs are different in artificial activity intensity, so that the optical active substances (such as chlorophyll a, colored soluble organic matters and suspended particles) of the water body are different and are reflected in the difference of the detection spectrum of the remote sensing sensor. In order to construct a universal chlorophyll a inversion algorithm model, lake water samples covering different lake regions across the country need to be obtained to construct training data sets of different water optical types. In 2011-2020, 65 lakes are investigated on site, water body samples are collected, and laboratory measurement is carried out to obtain 941 actual measurement chlorophyll a data of lakes and reservoirs, and the distribution of lakes and reservoirs is investigated and shown in figure 1.
The method is characterized in that the sampling points of each lake and reservoir are distributed at equal intervals, in the distribution process of the sampling points, the large lakes are distributed at a distance of 5km from the near bank of the lakes, and for the small lakes, the lakes are distributed according to the near condition of the lakes during actual sampling, so that the collected samples have certain representativeness.
The sample collection process is firstly kept clear and cloudy, and the collection of the water samples in the wild lakes is specifically divided into the following parts: it is first determined that the hull is to be moored to maintain the hull relatively stable. In the collection process, suspended substances floating on the water surface are avoided, a sample bottle (polyethylene) needs to be rinsed with hydrochloric acid in a laboratory before sampling, then a field water sample is used for rinsing, 1-L water body located between 0 and 0.5m of the water surface is collected, and the sample bottle is placed into a 4 ℃ vehicle-mounted refrigerator in a dark place and is subjected to laboratory analysis and test.
After the water sample is vibrated and shaken uniformly, the water sample passes through a peninsula mixed cellulose membrane with the aperture of 0.45 mu m under low vacuum pressure, the filter membrane is packaged by tinfoil after filtration, the mixture is frozen and preserved at the temperature of minus 20 ℃, cut into pieces and put into a 15mL centrifuge tube (protected from light and heat) for preservation, then 10mL 90% acetone (chromatographic pure) solution is added, the mixture is vibrated and kept still, and is extracted for 4 to 6 hours in a protected manner, and the longest time is not more than 12 hours. And (3) after the chlorophyll is fully extracted by the acetone solution, placing the chlorophyll into a centrifuge tube to be centrifuged for 20min in a centrifuge (rotating speed of 5000 rpm), taking supernatant after centrifugation, pouring the supernatant into a 1cm quartz cuvette, placing the cuvette in an Shimadzu UV-2600 ultraviolet visible light spectrophotometer to perform determination, and taking 90% acetone as blank reference. When measuring, the absorbance values at wavelengths of 750nm, 664nm, 647nm and 630nm were read, respectively, and the concentration value (unit, μ g/L) of Chl-a was calculated based on the Jeffrey-Humphrey equation and the volume of the filtered water sample
Rho is the mass concentration of chlorophyll a in the sample, mu g/L;
D 630 -absorbance value of the sample at a wavelength of 630 nm;
D 647 -absorbance value of the sample at a wavelength of 647 nm;
D 664 -absorbance value of the sample at a wavelength of 664 nm;
D 750 -absorbance value of the sample at a wavelength of 750 nm;
V 1 volumetric volume of sample, ml
V-sample volume, L
2. Remote sensing image data corresponding to a water sample collected on site are obtained based on a GEE computing cloud platform, then remote sensing image data of clouds, mountain shadows and atmospheric aerosol are removed, and effective matching data are screened out; extracting the reflectivity of the remote sensing zenith synchronously observed on site;
3. the image data of the Landsat series satellite TM/ETM +/OLI integrated by the GEE computing cloud platform is a zenith reflectivity product issued by the United States Geological Survey (USGS), the data set is the earliest remote sensing data at present, the revisiting period of a single terrestrial satellite is 16 days, and the spatial resolution is 30m. The method meets the spatial resolution requirement of the water quality parameter research of most inland water bodies worldwide, and has unique advantages for long-time sequence lake and reservoir water quality monitoring and climate change research. At present, the wave bands commonly used for water quality remote sensing inversion generally comprise four wave bands of blue wave band, green wave band, red wave band and near infrared. According to the on-site sampling date and position information of chlorophyll a in national lakes and reservoirs, by combining a GEE database and a geographical computing cloud, the zenith reflectivity and remark information corresponding to each on-site sampling position information are obtained, effective matching data are screened, and 141 landscapes Landsat satellite remote sensing images are downloaded, specifically:
3.1, the time window of remote sensing data acquisition is +/-7 days, the space window is 3 multiplied by 3 pixels, and the process can be realized through GEE cloud platform programming business.
3.2, in the process of acquiring the effective matching data, removing remote sensing image data which is greatly influenced by cloud, mountain shadow, atmospheric aerosol and the like according to remark information, and screening out the effective matching data.
3.3, in the process of effectively matching data acquisition, if the satellite images at the same time point comprise Landsat5 images and Landsat7 images, selecting the Landsat5 satellite images preferentially under the condition of similar cloud cover, wherein the phenomenon that partial data is lost due to the fact that the Landsat7 satellite images generate strips on the onboard scan line corrector in 2003 satellites;
3.4, band appears on ETM + DATA carried by Landsat-7 satellite in No. 31 of 5/2003, so that the DATA needs to be band-patched by calling a DEM _ BAD _ DATA _ DOIT command in IDL before using the ETM + DATA;
3.5, considering that the inversion of a long-time sequence needs to be carried out by the aid of three sensors of TM/ETM +/OLI at the same time, and in order to eliminate error information introduced by different sensors, linear fitting, namely consistency inspection, of the zenith reflectivity of each waveband needs to be carried out on the Landsat series satellite sensors. The remote sensing image data (table 3) of Tibet with relatively stable water body optical characteristics and a time window period of +/-1 day is selected, an ROI (region of interest) is drawn by means of an ROI (region of interest) tool of ENVI software, pixel values in the region of interest are exported to be ASCII (American standard code for information interchange) files, blue, green, red and near infrared wave bands are calculated to be the water surface remote sensing reflectivity, and a unary linear regression equation among the four wave bands of different sensors is established.
TABLE 3 quasi-synchronous cross-border image information of Tibet dislocation Hubei
4. The three-step effective matching data is obtained by randomly dividing the effective matching data into 70% chlorophyll a inversion model correction data sets by using a random function rand in Matlab software, N Correction of ＝2/3×N General assembly (ii) a 30% of the model validation data set, N Checking the right ＝1/3×N General assembly (ii) a Wherein, N General assembly Representing valid match data, N Correction of Representing chlorophyll a inverse model correction data set, N Checking the right Representing a chlorophyll a inverse model validation dataset.
5. Constructing a lake and reservoir chlorophyll a random forest inversion model, and carrying out high correlation test on zenith reflectivity values of all wave bands and actually-measured chlorophyll a aiming at a chlorophyll a inversion model correction data set, wherein a Landsat series satellite zenith reflectivity and actually-measured chlorophyll a concentration correlation coefficient distribution diagram is shown in figure 2. Preferably, a wave band with good correlation with chlorophyll a is used as an input variable of a random forest algorithm, and an inversion model is constructed, and the specific steps are as follows:
5.1, aiming at the model correction data set, analyzing the correlation between the zenith reflectivity value of each wave band and the actual measurement data of chlorophyll a in the first step, and selecting a wave band combination with the correlation; calculating the correlation (linearity, exponent and power function, etc.) between 58 wave band combination forms (ratio, difference, sum, etc.) and the measured chlorophyll a concentration by autonomous programming (table 2), and calculating the correlation coefficient;
TABLE 2 correlation test of band combinations
B denotes a wavelength band, G denotes a red wavelength band, R denotes a red wavelength band, and NIR denotes a red wavelength band.
5.2 correction of data sets for models (N) Correction of ＝2/3×N General assembly = 632), there are 16 band combinations having a band combination correlation coefficient of more than 0.6 (two tailors test, p)<0.001 The 16 wave band combinations with higher correlation are used as input variables of the random forest algorithm;
5.3 correction of data sets for models (N) Correction of ＝2/3×N General assembly = 632), constructing a model suitable for inverting the chlorophyll a random forest inversion in the lake and reservoir of China, performing fitting analysis on an inversion value and a measured value of the model, and fitting a linear regression coefficient R 2 =0.89, the root mean square error RMSE of the model correction data set is 0.465 μ g/L, the MAPE is 68.34 μ g/L, and a model inversion and actual measurement fitting graph of the random forest correction group of chlorophyll a in the water body of the lake and reservoir is shown in a figure 3;
5.4 validating the dataset against the model (N) Correction of ＝1/3×N General assembly = 309), carrying out precision verification on a chlorophyll a random forest inversion model, and carrying out model inversionFitting the values with the measured values to obtain a linear regression coefficient R 2 =0.75, the root mean square error RMSE of the model validation data set is 0.65 μ g/L, the MAPE is 39.04 μ g/L, and the accuracy of the international universal remote sensing inversion is met (<45%) of the water chlorophyll a random forest verification group model inversion and actual measurement fitting graphs in the lake and reservoir are shown in FIG. 4;
5. applying the chlorophyll a inversion model to the zenith reflectivity of the GEE storage remote sensing image, namely quickly drawing the chlorophyll a of the large-scale regional lake reservoir; the constructed model is applied to the Landsat series satellite data zenith reflectivity stored by the GEE by extracting the zenith reflectivity through autonomous programming by means of a GEE geographic computing cloud platform and an online data query function. Inversion of lake area in northern China in 1985-2020>1km 2 The 412 lakes and reservoirs in the lake and reservoir area take the chlorophyll change of lakes on multiple time scales (annual average, monthly average and climatic month) into consideration, and complete the standardized mapping of chlorophyll a of the lakes and reservoirs in large-scale areas.
Wherein, the lake area in the north of China in 1985-2020 is inverted>1km 2 412 lakes, (northern lakes in China in 1985-2020) ((>1km 2 ) Chlorophyll a concentration distribution graph is shown in FIG. 5-FIG. 12, wherein FIG. 5 is the lake in northern China (1985)>1km 2 ) Chlorophyll a concentration profile; FIG. 6 shows the lake reservoir in North China (1990)>1km 2 ) Chlorophyll a concentration profile; FIG. 7 is a 1995 northern lake repository in China (>1km 2 ) Chlorophyll a concentration profile; FIG. 8 shows the lake reservoir in North China in 2000 (>1km 2 ) Chlorophyll a concentration profile; FIG. 9 is 2005 northern lake repository in China (>1km 2 ) Chlorophyll a concentration profile; FIG. 10 shows the 2010 lake repository in northern China (>1km 2 ) Chlorophyll a concentration profile; FIG. 11 shows the lake reservoir in North China (2015)>1km 2 ) Chlorophyll a concentration profile; FIG. 12 is the lake reservoir in North China in 2020: (>1km 2 ) Chlorophyll a concentration profile.
As can be seen from figures 2 to 12, the method can accurately estimate the multi-space-time scale chlorophyll a concentration and the spatial variation condition monitoring thereof, in addition, the chlorophyll A remote sensing inversion model is constructed and obtained, and in practical application, the chlorophyll A remote sensing inversion model can be directly used without repeated data acquisition.
Claims (3)
1. A GEE-based large-scale region lake and reservoir chlorophyll a mapping method is characterized in that the GEE-based large-scale region lake and reservoir chlorophyll a mapping method is carried out according to the following steps:
1. collecting lake and reservoir water body samples with different climatic and geographical environment backgrounds to obtain actual measurement data of chlorophyll a;
2. remote sensing image data corresponding to a field-collected water sample are obtained based on a GEE computing cloud platform, and zenith reflectivity values and remark information of all wave bands are extracted;
3. removing remote sensing image data of clouds, mountain shadows and atmospheric aerosol according to remark information, and screening effective matching data;
4. randomly distributing the effective matching data obtained in the third step, wherein 70% of the effective matching data are chlorophyll a inversion model correction data sets, and 30% of the effective matching data are model verification data sets;
5. constructing a lake-reservoir chlorophyll a random forest inversion model based on chlorophyll a measured data, zenith reflectance values, a chlorophyll a inversion model correction data set and a chlorophyll a inversion model verification data set, and specifically comprising the following steps:
a) Aiming at the model correction data set, analyzing the correlation between the zenith reflectance value of each wave band and actually measured chlorophyll a data, and selecting a wave band with high correlation;
b) Aiming at the model correction data set, training different combination forms of wave bands with good correlation with chlorophyll a concentration, and using the different combination forms and the chlorophyll a concentration value as model input variables;
c) Parameterizing the model correction data set to construct a chlorophyll a remote sensing inversion model;
d) Verifying a chlorophyll a remote sensing inversion model aiming at the model verification data set;
6. and applying the chlorophyll a inversion model in the fifth step to the zenith reflectivity of the remote sensing image on the GEE platform to realize rapid drawing of chlorophyll a in the lake reservoir in a large-scale area and multiple space-time scales.
2. The method for mapping chlorophyll-a in large-scale region lakes and reservoirs based on GEE as claimed in claim 1, wherein the measured data of chlorophyll-a in step one is obtained by acetone extraction, i.e. calculating the concentration value of chlorophyll-a in μ g/L according to Jeffrey-Humphrey equation and the volume of filtered water sample.
3. The GEE-based large-scale regional lake reservoir chlorophyll a mapping method according to claim 1, characterized in that in step three, the random function rand in Matlab software is used to randomly divide effective matching data into 70% chlorophyll a inverse model correction data sets, N Correction of ＝2/3×N General assembly (ii) a 30% of the model validation data set, N Checking the right ＝1/3×N General (1) (ii) a Wherein N is General (1) Representing valid match data, N Correction of Data set of correction data representing chlorophyll a inverse model, N Checking the right Representing a chlorophyll a inverse model validation dataset.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210710949.5A CN115931783A (en) | 2022-06-22 | 2022-06-22 | GEE-based large-scale region lake and reservoir chlorophyll a mapping method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210710949.5A CN115931783A (en) | 2022-06-22 | 2022-06-22 | GEE-based large-scale region lake and reservoir chlorophyll a mapping method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115931783A true CN115931783A (en) | 2023-04-07 |
Family
ID=86651203
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210710949.5A Pending CN115931783A (en) | 2022-06-22 | 2022-06-22 | GEE-based large-scale region lake and reservoir chlorophyll a mapping method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115931783A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116451481A (en) * | 2023-04-19 | 2023-07-18 | 北京首创大气环境科技股份有限公司 | Multi-parameter rapid water quality inversion method based on GEE cloud platform and Sentinel-2 image |
-
2022
- 2022-06-22 CN CN202210710949.5A patent/CN115931783A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116451481A (en) * | 2023-04-19 | 2023-07-18 | 北京首创大气环境科技股份有限公司 | Multi-parameter rapid water quality inversion method based on GEE cloud platform and Sentinel-2 image |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Giardino et al. | Imaging spectrometry of inland and coastal waters: state of the art, achievements and perspectives | |
| Zoffoli et al. | Sentinel-2 remote sensing of Zostera noltei-dominated intertidal seagrass meadows | |
| Shuchman et al. | An algorithm to retrieve chlorophyll, dissolved organic carbon, and suspended minerals from Great Lakes satellite data | |
| Majozi et al. | Remote sensing of euphotic depth in shallow tropical inland waters of Lake Naivasha using MERIS data | |
| Brewin et al. | Underway spectrophotometry along the Atlantic Meridional Transect reveals high performance in satellite chlorophyll retrievals | |
| Dekker et al. | Retrospective seagrass change detection in a shallow coastal tidal Australian lake | |
| Jafar-Sidik et al. | The relationship between Suspended Particulate Matter and Turbidity at a mooring station in a coastal environment: consequences for satellite-derived products | |
| Heenkenda et al. | Quantifying mangrove chlorophyll from high spatial resolution imagery | |
| CN113203694B (en) | MSI lake eutrophication index remote sensing estimation method | |
| Kallio et al. | Validation of MERIS spectral inversion processors using reflectance, IOP and water quality measurements in boreal lakes | |
| Grendaitė et al. | Chlorophyll-a concentration retrieval in eutrophic lakes in Lithuania from Sentinel-2 data | |
| Fu et al. | Spatial downscaling of MODIS Chlorophyll-a using Landsat 8 images for complex coastal water monitoring | |
| CN114241331B (en) | Remote sensing modeling method for ground biomass of reed in wetland by taking UAV as ground and Septinel-2 medium | |
| CN110836870B (en) | GEE-based large-area lake transparency rapid drawing method | |
| Hang et al. | Estimation of chlorophyll-a concentration in Lake Taihu from Gaofen-1 wide-field-of-view data through a machine learning trained algorithm | |
| Liu et al. | Trophic state assessment of optically diverse lakes using Sentinel-3-derived trophic level index | |
| Mélin | Uncertainties in ocean colour remote sensing | |
| Chusnah et al. | Estimating chlorophyll-a concentrations in tropical reservoirs from band-ratio machine learning models | |
| CN115931783A (en) | GEE-based large-scale region lake and reservoir chlorophyll a mapping method | |
| Fan et al. | Machine learning algorithms for retrievals of aerosol and ocean color products from FY-3D MERSI-II instrument | |
| Danbara | Deriving water quality indicators of Lake Tana, Ethiopia, from Landsat-8 | |
| Rodríguez-López et al. | Spatio-temporal analysis of chlorophyll in six Araucanian lakes of Central-South Chile from Landsat imagery | |
| Mollaee | Estimation of phytoplankton chlorophyll-a concentration in the western basin of Lake Erie using Sentinel-2 and Sentinel-3 data | |
| Zhang et al. | Determination of phytoplankton abundances (Chlorophyll-a) in the optically complex inland water-The Baltic Sea | |
| Van Stokkom et al. | Quantitative use of passive optical remote sensing over coastal and inland water bodies |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |